I. Common Threats to generalization:

A. Interaction effects of settings

B. Interaction effects of stimuli and tasks

C. Interaction effects of participant selection bias

II. The distinction between generalizing across versus generalizing to

III. To generalize across different groups, settings, etc., use probability sampling of:

A. Settings

B. Stimuli and tasks

C. Participants

IV. To generalize to different groups, settings, etc.:

A. Replicate studies using different:

1. Settings

2. Stimuli and tasks

3. Participants

B. Look for statistical interactions between treatment and:

1. Settings

2. Stimuli and tasks

3. Participants

C. Template matching

V. Generalizing across time:

A. Statistical significance is no assurance of effect lasting over time, only means that results probably weren't due to chance.

B. The replication net

VI. Conclusions: If you try to generalize your results, you are making an inference. Confidence that the inference is correct may depend on:

A. Methodological considerations discussed above

B. Existing theories and laws about the phenomena studied

C. The extent to which previous research results in this field have been successfully generalized to other settings, participants, stimuli and tasks

Back to Chapter Main Menu